Setup

Data

To download into repository root:

• Sample_Program_Service_Data.csv - Download from Amazon Sample pre-aggregated financial and text data from Form 990s.

In repository:

• manula_datakind_labels.csv - ~345 hand-labeled SDG rows (about ten per SDG)
• NTEE_EINs_EOBMF.csv maps between EINs and NTEE taxonomy
• goals.txt contains SDGs and their descriptions

Code

This code requires Python 3 and pip. We recommend using virtual environments (via virtualenv or conda).

1. Run pip install -r requirements.txt to get all Python dependencies.

Analytics

Notebooks

• 990_analysis.ipynb - requires missing nonprofit-descriptions_2016.csv and NTEE_EINs.csv
• merged_data_models.ipynb - trains and evaluates a set of NLP models on the sample dataset, to predict SDGs

Other Notes

Producing data

Script to produce NTEE EINs from the IRS EOBMF

df = pd.read_csv("irs_eobmf.csv")
df = df[["ein", "ntee_cd"]].rename(columns={"ein": "EIN", "ntee_cd": "NTEE"})
df.to_csv("NTEE_EINs_EOBMF.csv", index=False)